add json

Author: litongmacos

resources/json/transcription01.json

@@ -0,0 +1,57 @@
+{
+  "transcription": [
+    {
+      "timestamps": {
+        "from": "00:00:00,000",
+        "to": "00:00:30,000"
+      },
+      "offsets": {
+        "from": 0,
+        "to": 30000
+      },
+      "token": {
+        "text": "[_BEG_]",
+        "id": 50363,
+        "confidence": 0.655302,
+        "t0": 0,
+        "t1": 0
+      },
+      "token": {
+        "text": " Thank",
+        "id": 6952,
+        "confidence": 0.673538,
+        "t0": 0,
+        "t1": 124
+      },
+      "token": {
+        "text": " you",
+        "id": 345,
+        "confidence": 0.931985,
+        "t0": 124,
+        "t1": 200
+      },
+      "token": {
+        "text": ".",
+        "id": 13,
+        "confidence": 0.826669,
+        "t0": 200,
+        "t1": 260
+      },
+      "token": {
+        "text": "[_TT_130]",
+        "id": 50493,
+        "confidence": 0.0699378,
+        "t0": 260,
+        "t1": 3000
+      },
+      "token": {
+        "text": "<|endoftext|>",
+        "id": 50256,
+        "confidence": 0.993674,
+        "t0": 3000,
+        "t1": 3000
+      },
+      "text": "[_BEG_] Thank you.[_TT_130]<|endoftext|>"
+    }
+  ]
+}

resources/json/whisper_local.json

@@ -0,0 +1,27 @@
+{
+  "systeminfo": "AVX = 0 | AVX2 = 0 | AVX512 = 0 | FMA = 0 | NEON = 1 | ARM_FMA = 1 | METAL = 1 | F16C = 0 | FP16_VA = 1 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 0 | SSSE3 = 0 | VSX = 0 | CUDA = 0 | COREML = 1 | OPENVINO = 0 | ",
+  "model": {
+    "type": "base",
+    "multilingual": false,
+    "vocab": 51864,
+    "audio": {
+      "ctx": 1500,
+      "state": 512,
+      "head": 8,
+      "layer": 6
+    },
+    "text": {
+      "ctx": 448,
+      "state": 512,
+      "head": 8,
+      "layer": 6
+    },
+    "mels": 80,
+    "ftype": 1
+  },
+  "params": {
+    "model": "models/ggml-base.en.bin",
+    "language": "en",
+    "translate": false
+  }
+}

stream_components_params.h

@@ -46,6 +46,7 @@ namespace stream_components {
 
     bool tinydiarize = false;
     bool diarize = false;
+    bool use_gpu = true;
 
     std::string language = "en";
     std::string model = "models/ggml-base.en.bin";

stream_components_service.cpp

@@ -5,12 +5,12 @@ using namespace stream_components;
 
 // -- WhisperService --
 
-WhisperService::WhisperService(
-  const struct service_params &server_params,
-  const struct audio_params &audio_params) :
-  server_params(server_params),
-  audio_params(audio_params),
-  ctx(whisper_init_from_file(server_params.model.c_str())) {
+WhisperService::WhisperService(const struct service_params &server_params,
+                               const struct audio_params &audio_params,
+                               const struct whisper_context_params &cparams)
+  : server_params(server_params),
+    audio_params(audio_params),
+    ctx(whisper_init_from_file_with_params(server_params.model.c_str(), cparams)) {
   {
     fprintf(stderr, "\n");
     if (!whisper_is_multilingual(ctx)) {

stream_components_service.h

@@ -1,5 +1,6 @@
 #ifndef WHISPER_STREAM_COMPONENTS_SERVER_H
 #define WHISPER_STREAM_COMPONENTS_SERVER_H
+
 #include <vector>
 #include <whisper.h>
 #include "stream_components_params.h"
@@ -12,11 +13,12 @@ namespace stream_components {
 /**
  * Encapsulates the Whisper service.
  */
-class WhisperService {
+  class WhisperService {
   public:
     WhisperService(
       const struct service_params &server_params,
-      const struct audio_params &audio_params);
+      const struct audio_params &audio_params,
+      const struct whisper_context_params &cparams);
 
     ~WhisperService();
 

stream_local.cpp

@@ -113,7 +113,9 @@ int main(int argc, char **argv) {
   stream_components::LocalSDLMicrophone microphone(params.audio);
 
   // Instantiate the service
-  stream_components::WhisperService whisperService(params.service, params.audio);
+  struct whisper_context_params cparams;
+  cparams.use_gpu = params.service.use_gpu;
+  stream_components::WhisperService whisperService(params.service, params.audio,cparams);
 
   // Print the 'header'...
   WhisperStreamOutput::to_json(std::cout, params.service, whisperService.ctx);
@@ -133,13 +135,11 @@ int main(int argc, char **argv) {
 
     // get the whisper output
     const WhisperOutputPtr &outputPtr = whisperService.process(pcmf32.data(), pcmf32.size());
-
     // write the output as json to stdout (for this example)
     if (outputPtr) {
       outputPtr->transcription_to_json(std::cout);
     }
   }
-
   std::cout << "EXITED MAIN LOOP" << std::endl;
   return 0;
 }